Cryptorhythmic and macrorhythmic layering in aegirine lujavrite, Ilfmaussaq alkaline intrusion, South Greenland*

J.C. BAILEY

Bailey, J. C. 1995: Cryptorhythmic and macrorhytl)mic layering in aegirine lujavrite, Ilfmaussaq alkaline intrusion, South Greenland. Bulletin of the Geological Society of Den­mark Denmark, vol. 42, pp.1-16. Copenhagen 1995-10-31. https://doi.org/10.37570/bgsd-1995-42-01

Cryptorhythmic layering (Wilson 1992) has been established in a 200 m drill core through a layered aegirine-eudialyte nepheline syenite, aegirine lujavrite I, of the Ilfmaussaq alkaline intrusion. Within each cryptorhythmic unit, the content ofY in cumulus eudialyte is cons­tant, but there is a sharp increase to a new fixed value in the next unit. Each cryptor­hythmic unit contains from two to four macrorhythmic units which show moderate to poor density sorting and consist of three layers defined by successively high contents of cumulus aegirine, then cumulus eudialyte, microcline and nepheline, and finally intercumulus zeolites and arfvedsonite. The "dry" density of both cumulus and intercumulus material decreases in the higher cryptorhythmic units and is thought to reflect the decreasing density of the parental liquid layers. The appearance of a new cryptorhythmic unit does not coincide with the end of a macrorhythmic unit and is apparently unrelated to pulses of nucleation and settling out of cumulus phases. Crystallization of cryptorhythmic units from chemically distinct double-diffusive liquid layers in the aegirine lujavrite I magma and of macrorhythmic units by repeated in situ crystallization on the floor of the Ilimaussaq magma chamber is discussed.

Bailey, J. C., Department of Petrology, Geological Institute, v}ster Voldgade 10, DK-1350 Copenhagen K, Denmark. November 25th, 1994.'

* Contribution to the Mineralogy of llfmaussaq no. 95.

Introduction

Igneous petrologists have long considered that frac­tional crystallization combined with the settling out of the fractionating phases was the dominant process inthe differentiation of magmas (Bowen 1928; Wager 1963). The process can be demonstrated on the mg­to-g scale in the laboratory (Bowen 1915) and in near­surface lava lakes (Helz 1987) but its recognition inthe geological record is largely a matter of interpreta­tion rather than unequivocal demonstration. Indeed, the settling out of crystals in magma chambers has recently been challenged and there is a body of opi­nion which considers it unlikely (Campbell 1978; McBirney & Noyes 1979; Sparks et al. 1984).

In the present article, the process of fractional crystal­lization is monitored through 200 m of a macrorhyth­mically layered nepheline syenite (aegirine lujavrite I) from the I](maussaq intrusion, South Greenland. Thelujavrite crystallized from a large syenitic pluton at alate stage in its evolution when the volume of the resi­dual magma chamber was small and when the residualmagma had reached an alkali-rich agpaitic, volatile­rich state. It is demonstrated that Y and U can be used

Bailey: Cryptorhythmic and macrorhythmic layering

to reveal systematic shifts in the chemistry of cumulus eudialyte but these shifts do not coincide with the boun­daries of macrorhythmic units.

Geological setting

The Ilfmaussaq intrusion (Fig. 1) forms part of a continental rift-zone province, the late Precambrian Gardar igneous province of South Greenland. Events began with a marginal intrusion of augite syenite and two thin sheets of peralkaline granite beneath the roof. The third, and most voluminous, pulse of magma was peralkaline and silica-undersaturated. It evolved to form a layered series of agpaitic nepheline syenites in the centre of the intrusion (Ussing 1912; Ferguson 1964; Bailey et al. 1981a; Larsen & S�rensen 1987).

The roof rocks of the third pulse - pulaskite, foyaite, sodalite foyaite and naujaite - crystallized successively downwards while contemporaneous bottom cumulates are thought to be present at depth. The exposed bottom cumulates, the kakortokites, contain blocks of the roof cumulates and are therefore younger. The kakortokites

46 ' I L I M M A A S A Q \\5'0 ; • ; • ; • • • ; • ; • ; • ; • ; • ; • • • ; • ; • ; • • ; ^ • ' _

GEOLOGICAL MAP OF THE ILIMAUSSAQ INTRUSION :: :

i l l Lujavrite, black i^ ^ l Lujavrite, green I I Kakortokite \>°«°«°.\ Nauja i te

\£l£Å Sodalite foyaite lllllllllll Pulaskite, foyaite \ttll\ Alkali grani te,

quartz syenite Augite syenite

liiiiilliiU Narssaq intrusion Gardar supracrustals Basement granite

Fig.l. Geological map of the Ilimaussaq intrusion after Ferguson (1964). The location of drill core 7 in aegirine lujavrite I is marked more accurately on Fig. 3.

2 Bulletin of the Geological Society of Denmark

augite syenite alkali acid rocks

pulaskite & foyaite sodalite foyaite

1500 m

naujaite

— / r->\-

1000

lujavrite

500

kakortokite

medium- to coarse- grained

lujavrite

arfvedsonite

lujavrite

lujavrite ~--̂ - ~'5S transition zone ^ S ^ ^ l p i l ^ S S S

aegirine lujavriten

aegirine lujavrite I

transitional layered k.

slightly layered k.

lower

layered

kakortokite

Fig.2. A. Schematic section through the Ilfmaussaq intrusion. B. The lowermost 800 m of the intrusion showing the detailed igneous stratigraphy of the kakortokites and lujavrites. C. ZrO profile for drill core 7 in aegirine lujavrite I. High Zr02 contents correspond to eudialyte-rich cumulate layers. After Andersen et al. (1981a) and Bailey et al. (1981b).

100 m 200

-v^r • ^ H s * , . « ^

Naujaite, roof rock and xenoliths

35S531 Aegirine lujavrite I with '-̂ -JS-BJ eudialyte-rich layers

i •¥"&:JS?J M e t a s o m a t i s e d

'&!?'••$ n a u i a i t e

T r a n s i t i o n a l

k a k o r t o k i t e

layered f a u l t

-* - d ip

A he igh t , m

(•) d r i l l co re

Fig. 3. Outcrop map of aegirine lujavrite I in the southeastern part of the Ilfmaussaq intrusion (Fig. 1) from Henriksen (1993). Only the more conspicuous eudialyte-rich layers can be followed in the field and more have been recognised by geochemical and modal logging of drill core 7.

Bailey: Cryptorhythmic and macrorhythmic layering 3

modal % -> wt % 'dry' densities (g/cm3) „ „ „ ^ ; „ a ~ J ; „ I t a microcline analcime , . .. ,. _. eud m+n macro-(»)and c r yp to -O aegirme eudialyte + n e phel ine +natrolite arfvedsonite K 2 0 a e g _ z + a rhythmic units

0 20 0 20 30 20 50 2-70 2-76 2-82

T

5 Is r

v

7^

J 5 L z

_£_ s

cv~

X

~7^~

J ^ z x

i

y ^z

M7

M10

M7

M6,

M5

M4n

M3

M2

M1

M0

M9n []

M8 7u

macrorhythmic units with individual layers

C3 ©

—F

C4 O

C3 O

C2 O

C1 O

CO O

cryptorhythmic units forY forU

Fig. 4. Modal (vol.%) and K20 (wt.%) analyses, and macrorhythmic and cryptorhythmic layers, for drill core 7, aegirine lujavrite I, Ih'maussaq intrusion. Aeg - aegirine, eud - eudialyte, m+n microcline and nepheline, z+a zeolites and arfvedsonite; F inferred fault. Proposed boundaries and internal layering of 10 macrorhythmic layers also incorporate data on Zr02 contents (Fig. 2). "Dry" densities of macro- and cryptorhythmic layers are based on modal analyses and measured mineral densities. Density symbols with arrows refer to incomplete layers where the overall density is expected to be higher than the measured. Note that the boundaries between cryptorhythmic layers as defined by the Y and U contents of eudialyte only coincide in one instance, at 152 m depth.

4 Bulletin of the Geological Society of Denmark

range through well layered, weakly layered and tran­sitional layered varieties before passing conformably upwards into lujavrites which have been divided into several varieties (Fig. 2B). The transition between the kakortokites and lujavrites probably reflects physical changes in a shrinking magma chamber due to subsi­dence and fracturing of the roof, causing lujavrites to evolve independently in two or more sub-chambers and to locally intrude and brecciate the overlying naujaite roof.

The earliest of the lujavrite varieties is an aegirine-rich nepheline syenite orthocumulate termed aegirine lujavrite I (Andersen et al. 1981a; Bohse & Andersen 1981; Bailey & Gwozdz 1994). It conformably overlies the transitional layered kakortokite in the southeastern part of the intrusion and extends for 4 km along the Lakseelv valley at the head of Kangerluarsuk fjord. It forms a sheet-like body thinning out towards the south where the naujaite roof extends to lower levels. The sheet ranges from about 50-300 m in thickness and averages around 80 m.

Drill core 7 (GGU number 108107) is located 1.0 km NE of the mouth of Lakseelv at an altitude of 120 m and was obtained from the southern end of aegirine lujavrite I near its middle levels (Figs. 1,2C, 3). It did not penetrate the underlying kakortokites. Xenoliths of the overlying naujaite roof occupy about 10% of the core. Since the lujavrite dips at 45 +/- 10° both in the surface outcrops and within most of the core (assumed to be vertical), the 200 m of core probably represents a true stratigraphic thickness close to 140 m. However, the uppermost 18 m of the core is a faulted repetition of deeper levels. The present dip of the core can be related to tilting associated with nearby faults, and the slope of the original depositional floor is unknown.

Analytical methods Drill core 7 was divided into 1 m intervals which were crushed in an agate mortar for 30 minutes. Zr and Y were determined by X-ray fluorescence spectrometry (XRFS) on powder pellets using the techniques of Norrish & Chappell (1977). Interference from Sr Koc and Rb K(3 on Zr Koc and Y Ka, respectively, was corrected for. Matrix variation was monitored by measuring the Mo Ka Compton peak in an off-peak position (-0.15 °20 using a LiF (220) crystal) to avoid interferences from Nb and Y lines.

Precision for Zr and Y was inside 1% relative and results were standardised against USGS reference materials (G-2, GSP-1, AGV-1, W-l and BCR-1) using the recommended values of Govindaraju (1994). Deter­mined (and recommended) values in ppm for the in­ternational lujavrite standard NIM-L are Zr 10600 (11000), Y 20.1 (22).

U was analysed by delayed neutron analysis at the Risø National Laboratory. Imprecision typically ran­

ges from +/- 3% at the 10 ppm level to +/- 2% at the 40 ppm level.

The remaining elements in this report were deter­mined by instrumental neutron activation (REE, Cs, Th, Hf, Ta, Co, Sc), a combined neutron activation and Cerenkov counting technique (Li, F), and a variety of XRFS methods.

Mineral separation was effected by a combination of heavy liquid and magnetic techniques. Impurities (0.5-4 vol.%) were visually estimated and analyses adjusted accordingly.

Modal analyses are based on counting about 500 points in one thin section per metre. Bulk "dry" densities of rocks are based on modal analyses and on mineral densities which were determined on calibrated mixtures of heavy liquids.

Macrorhythmic layering Macrorhythmic layering (Irvine 1987) in aegirine luja­vrite I was first recognised in drill core 7 on the basis of Zr (eudialyte) contents coupled with hand specimen and microscope observations (Bailey et al. 1981b; Bailey & Gwozdz 1994) and was later confirmed by detailed mapping (Henriksen 1993). It is now reap­praised using modal analyses and contents of Kft as a monitor of microcline abundances.

Unlike the spectacular macrorhythmic layering of the lower layered kakortokites (Ussing 1912; Søren­sen 1969; Ferguson 1970; Bohse et al. 1971), aegirine lujavrite I only exhibits moderately to poorly density-stratified macrorhythmic units. These consist of three individual layers but, unlike the kakortokites, they are successively rich in cumulus aegirine, then cumulus eudialyte, microcline and nepheline, and finally inter-cumulus material (zeolites and arfvedsonite) (Fig. 4).

Allowing for the dip in drill core 7, individual ma­crorhythmic units have real thicknesses of 6-17 m; their three constituent layers grade into each other and are typically 3-8 m thick. The boundary to the next macror­hythmic unit is always gradual, commonly over se­veral decimetres.

All three individual layers show planar lamination due to alignment of the platy microcline and acicular aegirine in the plane of layering. Although the upper­most layer crystallised from material rich in inter-cumulus liquid, it is still a cumulate rock in terms of its fabric and orthocumulate texture. Signs of current action such as scoured out troughs and current bed­ding have not been observed, and cumulus crystals are not linearly aligned.

The macrorhythmic units have been labelled M0 to M10 (Fig. 4). Unit M0, at the bottom of drill core 7, is incompletely represented. Unit Ml includes a very eudialyte-rich layer whose red colour can be easily fol­lowed in the field. Samples from 0-18 m depth, on the basis of layer thickness, modal contents and trends,

Bailey: Cryptorhythmic and macrorhythmic layering 5

Table 1. Whole-rock and mineral separate analyses, aegirine lujavrite I, sample 146 m.

Si02

Ti02

Zr02

Nbp5

^2°, AlA Fep3 FeO MnO MgO CaO Na,0 KO L ip p2°s H 2 0

Hp+

F Sum

Zr(ppm) Y U

Whole-rock

51.69 0.25 1.48 0.15 0.59

10.92 12.05 2.08 0.25 0.01 1.35

11.01 2.80 0.032 0.02 0.31 3.70 0.06

98.73

10800 964

23.4

Aegirine

52.43 0.43 0.24 0.02 0.05 0.28

30.04 1.35 0.07 0.02 0.96

12.67 0.01 0.007

<0.01 0.414

-<0.05 99.09

2500 19 6.2

Eudialyte

50.56 0.0

13.63 1.12 4.85 0.23 -2.383

1.64 0.02 8.82

12.72 1.00 0.002 ---0.05

97.07

100900 5750

143

M + N1

63.97 0.03 0.05 0.01 0.03

18.38 0.15 0.06 0.05 0.01 0.04 0.65

15.85 0.001

<0.01 0.294

-<0.05 99.54

130 31

3.5

Arfvedsonite

50.64 0.43 0.06 0.02 0.03 0.43

13.02 20.67

1.02 0.03 0.34 8.13 2.56 0.429

<0.01 1.794

-0.28

99.90

570 8 3.7

A + N2

47.09 0.03 0.06 0.02 0.03

25.97 0.33 0.02 0.10

<0.01 0.24

15.62 0.66 0.007

<0.01 9.164

-0.11

99.41

160 15 5.0

Analysts: I. Sørensen, H. Bohse, J.C. Bailey, R. Gwozdz. 1 microcline + nepheline; 2 analcime + natrolite; 3 total Fe as FeO; 4 LOI. All elements by XRFS except Li and F by INA-CC, U by DN, FeO by the metavanadate technique, Nap by atomic absorption, and H p and LOI by gravimetry.

bulk density and the Y-U chemistry of eudialyte (see below), are interpreted to represent a faulted repeti­tion of unit M7. Although the upper boundary of the underlying M10 unit has been faulted, M10 seems to be virtually complete.

Since several eudialyte-rich layers can be traced over an area of at least 1.3 x 0.3 km (Fig. 3) (Henriksen 1993), it seems likely that the macrorhythmic units established in core 7 have a considerable lateral extent.

The decision to locate the start of a macrorhythmic unit at the onset of an aegirine layer carries with it the important consequence that each macrorhythmic unit shows a rough density sorting with an aegirine layer (s.g. around 2.80 g.cm3) followed by a eudialyte and microcline + nepheline layer (c. 2.75) and a layer rich in intercumulus material (c. 2.73). The range of den­sity values is smaller than the range for the black (c. 3.07), red (c. 2.79) and white (c. 2.77) layers of the kakortokites (Ferguson 1970). Furthermore, in aegirine lujavrite I, the eudialyte layers show no clear signs of being located below the microcline + nepheline layers (Fig. 4). Thus density sorting can only be characterized as moderate to poor.

Bulk densities of the 10 complete macrorhythmic units show an overall decrease towards the higher units (r = 0.59) (Fig. 4). This is due to (a) the falling propor­

tion of cumulus phases in the upper units (r = 0.62) especially the two densest - aegirine and eudialyte (r = 0.67), (b) the decreasing density of the bulk cumulus phases (r = 0.21) and (c) the decreasing density of the intercumulus material (r = 0.37) due to increasing amounts of zeolites (r =- 0.65) towards the higher units. Post-magmatic zeolitization partly explains some of these features but the same trends have also been noted in the least zeolitized samples (Bailey & Gwozdz 1994).

Petrography Aegirine lujavrite I shows a marked textural and mine-ralogical distinction between the cumulus phases (aegi­rine, eudialyte, microcline, nepheline) and the intercu­mulus material (arfvedsonite, natrolite, analcime) (Bailey & Gwozdz 1994). Aegirine needles, microcline plates and the nearly equant eudialyte and nepheline crystals are all euhedral to subhedral; they are set in a matrix of interstitial to subpoikilitic natrolite and analcime together with ovoids (oikocrysts) of arfved­sonite. Primary igneous minerals and textures are commonly replaced by the matrix zeolites. The most

6 Bulletin of the Geological Society of Denmark

Table 2. Analyses of eudialytes from aegirine lujavrite I and an enclosed naujaite xenolith, Ilimaussaq intrusion.

Sample

Rock

Si02 Ti02 ZrO, Hf02 Nb205 Ta205 RE203

A I 22 O 3

3

FeO* MnO MgO CaO Na20 Kp F Cl Sum Li(ppm) Rb Sr Pb Ba La Ce Pr Nd Sm Gd Dy Ho Er Yb Lu Y Th U Zn La/YbN Th/U

D7 146 m aeg luj I

50.56 0.05

13.63 0.18 1.12 0.04 4.12 0.23 2.38 1.64 0.02 8.82

12.72 1.00 0.05 0.94

97.50 8.5

281 695 778 905

6210 14600

1930 6440 1330 1370 1300 282 756 704 100

5750 65

143 699

5.82 0.45

D7 96 m aeg luj I

49.86 0.05

13.34 0.16 1.14 0.04 4.42 0.34 3.05 1.88 0.03 8.37

13.19 0.65

<0.05 0.87

97.39 7.5

116 749 197 968

6880 16300 2170 7240 1510 1530 1430 296 826 759 106

6410 85

174 93

6.00 0.49

D7 46 m aeg luj I

49.06 0.03

13.79 0.17 1.24 0.04 4.87 0.48 3.16 1.77 0.07 8.03

12.66 1.02 0.12 0.83

97.34 13

147 702 666 790

7540 17600 2280 7660 1530 1550 1410 286 796 734 101

6410 74

156 993

6.78 0.47

D7 4 m aeg luj I

50.13 0.06

13.03 0.17 1.08 0.04 4.10 0.41 3.74 1.65 0.02 8.37

13.00 0.61 0.09 0.87

97.37 32

172 694 871

1100 6090

14500 1930 6460 1340 1390 1300 272 758 698 94

5770 88

154 1060

5.76 0.57

D7 33.7 m naujaite

49.01 0.07

13.29 0.23 1.09 0.06 1.86 0.68 6.08 1.07 0.03

10.48 13.24 0.25

<0.05 1.33

98.77 17 39

476 246 913

3010 6560

806 2620

506 586 644 156 434 417

61 2810

35 52

130 4.77 0.67

Analysts: H. Bohse, J.C. Bailey, R. Gwozdz. All elements were determined by XRFS except for Li and F by INA-CC. Low sums are due to non-determination of H.O.

common accessory phases are acmite, albite, sphalerite, native lead and iron oxides, whereas Li-mica (polyli-thionite) and molybdenite are only occasionally visi­ble. Naujaite xenoliths, sometimes intensely altered, have contaminated adjacent lujavrites in a variety of ways. Patches and veins of zeolites together with shear, fracture and breccia zones commonly with later pow­dery coatings can be assigned to post-magmatic stages.

Sub- to euhedral eudialyte crystals, around 1 mm in diameter, are present at a background level of 2-5 vol.% in the core but some 11 eudialy te-rich layers containing maxima of 10-35 vol.% eudialyte also occur. The eudia­lyte crystals may include a few percent of aegirine

needles up to 0.1 mm in length, that is about 10 times smaller than the cumulus aegirine crystals of the rock. Most eudialytes are optically homogeneous but occa­sionally one or two oscillatory zones may be present towards the outer margin. More rarely it is possible to observe patchy zoning of cores, non-concentric zoning, a single marginal zone or weak sector zoning. Over­growths, amounting to about 5% of an individual crystal, may develop at one of the sides or corners of a eudialyte crystal but are also uncommon. In zeolitised samples, eudialyte grains are commonly strained or weakly to intensely cracked with slight dispersal of the fragments, and show alteration patches of natrolite,

Bailey: Cryptorhythmic and macrorhythmic layering 7

Table 3. Analyses of coexisting minerals in two samples of aegirine lujavrite I.

Sample

Mineral

96 m aeg

191 m aeg

96 m eud

191 m eud

96 m m+n

191 m m+n

96 m arf

191 m arf

96 m a+n

191 m a+n

Cs Rb K% Li Ca% Sr Pb La Ce Nd Sm Eu Tb Yb Lu Y Th U Zr% Hf Ta Mg% Co Zn Sc Ga K/Rb La/YbN

Th/U Zr/Hf

<0.3 4 0.87 16 0.68 6 96 52.6 118 -9.1 0.58 -5.35 0.90 41 6.9 5.6 0.21

55.3 2.0 0.04

<0.3 172 0.3 56 220 6.5 1.2

38

<0.8 3 0.11 10 0.92 6

275 32.3 74.4 -6.2 0.47 -3.31 0.78 28 14.6 3.9 0.25

62.6 1.0 0.02

<0.2 385 0.3

54 . 370 6.4 3.7 39

<3 47 0.35 5.0 6.18

777 180

6590 16300 7500 1540 141 218 686 103

6700 100 174

9.87 1410 370

0.02 <1.0 31 <0.1 <1 75 6.3 0.6 70

<3 -0.38 8.6 7.34

654 990

6530 15600 7030 1560 129 219 707 110

7140 127 130 10.21

1760 420

0.03 <0.9 102 <0.1 <1 -6.1 1.0 58

2.6 3110

13.2 8.2 0.01 10 260 20 47 -2.2 0.11 0.52 1.6 0.25 20 9.4 10.4 0.01 5.3 1.7 0.01

<0.2 43 <0.1 76 42 8.3 0.9 24

1.8 2570

12.5 5.9 0.04 11 160 30 63 -2.9 0.15 0.55 1.9 0.29 22 8.4 2.6 0.01 3.1 0.5

<0.01 <2.9 119 <0.1 70 49 10.4 3.2 38

<0.9 79 2.15

2090 0.24 17 40 12.2 33.7 -1.9 --3.16 0.68 29 0.54 2.8 0.07 18.4 1.5 0.04 1.6

553 <0.1 33 270 2.5 0.2 39

1.7 104

2.24 2020

0.28 14 600 24.7 55.7 -3.6 --4.77 0.85 29 4.0 3.9 0.07

21.4 1.2 0.04 1.2

540 0.4 32 215 3.4 1.0 31

37 114 0.41 21 0.16 8 23 37.3 81.6 -5.2 0.25 0.59 1.67 0.22 23 3.1 2.6 0.01 2.7 1.2 0.02

<3.0 57 <0.1 61 36 14.7 1.2 41

42 128 0.48 32 0.28 10 718 26.7 65.2 -4.8 0.36 0.68 1.16 0.14 16 11.6 3.2 0.02 5.6 0.8

<0.01 <0.1 250 <0.1 46 31 15.2 3.6 43

Analysts: R. Gwozdz, H. Kunzendorf, J.C. Bailey. Minerals: aeg - aegirine; eud - eudialyte; m+n - microcline + nepheline; arf - arfvedsonite; a+n - analcime + natrolite. All elements were determined by INA except for Rb, K, Ca, Sr, Pb, Y, Zr, Mg, Zn and Ga by XRFS, Li by INA-CC and U by DN.

analcime, a Na-Zr-REE phase (?catapleiite) and zirfe-site.

Mineral chemistry Analyses of rock-forming minerals in aegirine lujavrite I are characterized by low-temperature end-member compositions (Table 1). Thus MgO contents in aegirine and arfvedsonite are at trace-element levels, microcline is virtually Na20-free and eudialyte is the end-member of the eucolite-mesodialyte-eudialyte series (Vlasov 1966). In the absence of major element substitutions, normal methods of checking cryptic variations in an igneous sequence using electron microprobe techniques are unsuitable.

However, cryptic variations are present among trace elements, notably REE, Th, U, Hf and Nb in separated

eudialytes (Table 2). Contents of these elements in eudialyte increase in the higher layers of aegirine lujavrite, though there is a reversal in the sample from the highest level at 4 m depth in the core where faulting has repeated the igneous stratigraphy (see below).

Additional trace element analyses on separated mi­nerals are presented in Table 3 and indicate that the bulk of REE, Th, U, Zr, Hf, Nb and Ta is located in eudialyte. Preliminary work thus suggests that cryptic changes can be followed by analysing these elements in eudialyte but that the changes are only around a few percent relative and would be difficult to monitor by microprobe or neutron activation analytical tech­niques. XRFS analyses have the required precision but it was found that mineral separates, even of high purity, are not fully representative of a mineral phase because of selective losses during the separation procedures, e.g. crushing and removal of rims and alteration areas from eudialytes. However, at the n00-n000 ppm levels

Bulletin of the Geological Society of Denmark

found for several of these elements in these modally variable rocks, it is possible to perform whole-rock ana­lyses with high precision (see Analytical Methods) and to infer mineral chemistry with a high degree of confidence.

Whole-rock analyses of Zr, Y and U Contents of Zr, Y and U in one metre samples from drill core 7 through aegirine lujavrite I are plotted in Fig. 5. There is a clear coherence between the three elements due to their marked concentration in eudialyte and thus in eudialyte-rich layers. There are no obvious, overall trends for Zr and Y through the drill core but U maxima and minima tend to increase upwards from 30 to 40 ppm and 12 to 15 ppm, respectively.

Eudialyte-rich layers have higher Y/Zr but lower U/ Zr ratios than other layers and this causes these ratios to vary with the abundance of eudialyte (not shown). Overall, these ratios increase in their maximum and minimum values through the drill core suggesting that the higher eudialytes are richer in Y and U.

Plots of Zr-Y and Zr-U (Fig. 6) indicate that a num­ber of linear arrays are present. Each array is restricted to a specific depth interval in the drill core and the various arrays are interpreted as step-like shifts in the chemistry of eudialyte. Each line unites samples with fixed contents of Y and U in eudialyte, and the lines can be extrapolated to the Zr content of eudialyte (10.0 wt.%) in order to calculate these contents (Table 4).

The inferred Y contents (9900 to 11550 ppm) and U contents (118 to 210 ppm) of eudialytes increase at higher stratigraphic levels. For both elements, contents obtained by extrapolation generally exceed those found by analysis of separated eudialytes (Table 2). This fea­ture was noted by Bailey et al. (1983) and Bailey & Gwozdz (1994), and can be attributed to loss of Y and U during mineral separation when the easily crushed, altered grains of eudialyte and the U- (and probably) Y-rich rims of eudialytes tend to be lost. It is important to emphasise, however, that within whole-rock samples such Y and U must have been retained in situ through­out crystallization and (in many samples) during post-magmatic zeolitization; only by such retention can one explain the highly regular Zr-Y and Zr-U arrays of Fig. 6.

En masse, the remaining rock-forming minerals also show step-like shifts in their Y and U contents. At the Zr content of eudialyte-free lujavrite (2500 ppm), and according to the Zr-Y arrays, the Y contents of eudialyte-free lujavrite increase in a series of steps to­wards the higher stratigraphic levels of the core (Table 4). The same conclusion is reached for U though the innate scatter of the Zr-U arrays adds an air of uncer­tainty.

The percentage standard deviation of Zr and Y contents about each Zr-Y regression line ranges from about 1% at high Zr-Y levels to 2% at lower levels

and can thus be explained by the combined imprecision of the Zr and Y analyses (see Analytical Methods).

Only four samples out of 179 lie more than 36 away from their inferred Zr-Y regression lines and these deviations are mainly explained by post-magmatic al­teration. Zr-U relations are less regular. In order to establish the Zr-U regression lines of Fig. 6 it was necessary to reject 32 out of 179 samples. This proce­dure was justified by (a) the reasonable degree of Zr-U coherence displayed by the remaining 147 samples and (b) the clear linkage found between the deviant samples and the occurrence of post-magmatic zeolitization. For example, zeolitized samples in the sequence from 125.5-129.5 m have all lost some U.

Only the implications of the regular Zr-Y and Zr-U relations are considered further.

The levels of step-like shifts for Y and U do not corre­spond to the junctions between macrorhythmic units. Instead, the Y steps established at 174, 152, 122 and 88 m depth all lie in eudialyte-rich layers, being located in the lower, middle, middle and upper parts of these layers, respectively. The smooth rise and fall of Zr and the modal trends in these layers show no sign of any deviation at the step-like boundaries. Furthermore, most eudialyte layers are not associated with these boundaries.

In only a single example does the position of an Y step coincide with the position of a U step. This is at 152 m depth where the Y step is reliably established to 152 +/- 0.5 m and the U step is less securely established to 152 +/- 2 m.

Steps in the Y and U contents of eudialyte are all positive. The only apparent reversal is in the uppermost 18 m of the drill core. Here, Y contents of eudialytes are indistinguishable from those at 89.5-121.5 m depth (Fig. 6) and U contents can be matched with those from 18.5-105 m depth. Thus, the uppermost 18 m of the drill core corresponds to stratigraphic levels between 89.5 and 105 m, a conclusion already reached on the basis of modal variations and attributed to re­petition by faulting (see above). Although two faults parallel to the lujavrite layering have been recognised just south of drill core 7 (Fig. 3), smaller faults are difficult to recognise in these fissile lujavrites and, at the moment, field control is inadequate to confirm or deny the inferred stratigraphic repetition.

The conclusion that Y and U contents in eudialyte can be repeated by faulting, and can be integrated with a similar faulted repetition of the macrorhythmic layering, suggests that the steps in Y and U contents are also stratigraphic features and define the boundaries of stratified compositional units. On the basis of steps in the Y contents of eudialyte, these units have been labelled CO to C4 (Fig. 4).

The steps in Y contents of eudialyte are generally around 4% relative, and are roughly eight times less pronounced than those for U. There is a weak tendency (r = 0.34) for the Y steps to be greater at higher levels in the core and a similar tendency seems to occur for

Bailey: Cryptorhythmic and macrorhythmic layering 9

Table 4. Cryptorhythmic units as defined by contents of Y and U in cumulus eudialyte, aegirine lujavrite I, Ilimaussaq intrusion.

Cryptorhythmic unit C3 C4 C3 C2 CI CO

Depth (m) 0.5-17.5 Real thickness (m) >13 Y in eudialyte (ppm)1 11100 Y in eudialyte (ppm)2 7790

Y at 2500 ppm Zr1 107

Depth (m) 0.5-17.5 U in eudialyte (ppm)1 210 U in eudialyte (ppm)2 154

U at 2500 ppm Zr1 11.4

8.5-87.5 48

1550 8000 8200 152

8.5-105 210

52, 156 174 11.4

89.5-121.5 22

11100 7690

107

105-150 152 143

10.7

122.5-151.5 20

10700 7380

88

151.5-164 1153

-

11.23

152.5-173.5 15

10200 -

81

180-197.5 118 130

8.4

174.5->16

9900 7140

72

-197.5

See Fig. 4 and text for discussion. 1 - by extrapolation from Zr-Y and Zr-U arrays (Fig. 6); 2 - XRFS (Y) and DN (U) analysis of individual separated eudialytes; 3 - unreliable values because of limited points on Zr-U array.

the U steps also. At the moment, however, the number of steps available for checking this trend is barely adequate.

The sizes of the chemical steps appear to be correlated with the thicknesses of the stratigraphic units they define. Thus, the lowest unit for Y where bottom and top are both defined is only 15 m thick, but the overlying units are successively 20, 22 and at least 48 m thick. The top of this last unit is thought to be terminated by a fault and may originally have been greater. Against this apparent regularity, however, one should note that below the lowest defined unit there is an incomplete unit which is at least 16 m thick. The thicknesses of U units are less reliably defined but also seem to increase towards the higher stratigraphic levels of the core.

Setting on one side the incomplete Y unit CO, it is found that from unit CI to C4 there is a steady decrease (r = 0.88) in the bulk "dry" density from 2.828 to 2.727 g.cnr3 (Fig. 4). Since the density of both cumulus (r = 0.94) and intercumulus phases (r = 0.34) in units CI to C4 decreases, it seems reasonable to infer that the densities of the parental magmas for the Y units were also decreasing.

Discussion

Cryptorhythmic layering Wilson (1992) informally proposed the term "cryp­torhythmic layering" to describe the presence of dis­crete geochemical arrays over specific stratigraphic intervals in cumulus bronzitites of the layered Great Dyke, Zimbabwe. Within a single array, an incom­patible element such as Ti increases with differentiation

of the bronzite, i.e. with falling Mg/Mg + Fe, reflec­ting its uptake during the reaction between adcumulus bronzite and trapped intercumulus liquid. Other arrays show the same trend but at different absolute contents of Ti, and thus point to compositional discontinuities in the magma chamber. It was suggested that the se­ries of cryptorhythmic units developed by crystalliza­tion from a series of double-diffusive convection layers in the Great Dyke magma chamber.

The step-like changes in the chemistry of eudialyte and, en masse, in the other rock-forming minerals in aegirine lujavrite I satisfy the criteria for crypto­rhythmic layering and are also thought to reflect crystallization from a series of compositionally distinct liquid layers.

A more rigorous linkage between the chemistry of eudialytes and coexisting magmatic liquids would require analysis of eudialyte cores only. In the technique used here, the inferred contents of Y in eudialyte are bulk contents for cores and rims as well as rare over­growths and alteration patches. The geochemical regu­larity within an individual array apparently demands that both the overall crystallization history of eudialytes within the array and the composition of the coexisting magmatic liquid were constant.

The simplest explanation for the step-like shift to the next array is that there was also a step-like shift in magma composition. A more complex alternative is to argue that all the lujavrite magmas and coexisting eudialyte cores represented by drill core 7 were cons­tant in composition but that the subsequent crystal­lization history of eudialyte rims etc. had a step-like character. This would imply that amounts and/or compositions of intercumulus and/or post-magmatic liquids shifted in a step-like manner. Petrographic evidence of such steps, however, has not been found.

10 Bulletin of the Geological Society of Denmark

o

20

40-

Zr ppm 8000 16000 0

Y ppm 800 1600 0

T

"s:

i~^ T

'al >

-J J _ L

U ppm 20 40

_ 1 I L_

Fig. 5. Zr, Y and U contents in drill core 7, aegirine lujavrite I. Gaps in the sequence are occupied by xenoliths from the naujaite roof rocks.

Bailey: Cryptorhythmic and macrorhythmic layering 11

Furthermore, step-like changes in the amount of inter­cumulus magma and of final residues of this magma (Bailey & Gwozdz 1994) show no relation to the cryp­torhythmic boundaries established here.

Unlike the bronzites of the Great Dyke (Wilson 1992), the levels of Y in the lujavritic eudialytes are constant within a single cryptorhythmic unit. In part, this is thought to reflect the orthocumulate character of aegirine lujavrite I and the lack of interaction between the cumulus eudialyte and the intercumulus liquid, but it also points to the absence of fractionation processes within a cryptorhythmic unit.

The thicknesses of the best defined cryptorhythmic units in the Great Dyke range from 3-30 m. Units in aegirine lujavrite I tend to be thicker, ranging from 15->48 m for Y and from roughly 9—>61 m for U.

The presence of liquid layers at the aegirine lujavrite I stage of the Ilimaussaq magma chamber receives support from the decreasing bulk "dry" density of the rocks in the higher cryptorhythmic units. It is agreed that the only stable arrangement for liquid layers in a magma chamber is one of decreasing density at higher levels and that density differences of <1% (at least in simple aqueous systems) can be maintained by dou­ble-diffusive convection (Turner and Gustafson 1981). Differences of 0.5-2.4% rel. are observed in the "dry" rock density of adjacent layers in drill core 7.

The decrease in aegirine contents -21.4 to 14.6 vol.% equivalent to a decrease from 28.3 to 19.9 wt.% and a density decrease of 0.302 g.cnr3 - more than explains the decrease in rock densities of the cryptorhythmic units from 2.828 to 2.727 g.cnr3. The sharp density decrease reflecting the fall in aegirine abundance is mainly compensated by the rising proportion of intercu­mulus material, especially the zeolites. These changes in mineral proportions point to decreasing levels of Fe and increasing H20 in successive liquid layers.

Macrorhythmic layering One model of fractionation in magma chambers (Wager and Brown 1968) considers that dense cumulus phases nucleate near the cold roof of a chamber, and then sink or are carried by convection currents to the chamber floor. On this model, density-sorted macrorhythmic units represent the arrival of a pulse of crystals and should be integrated with cryptic changes in the cumulus phases.

In aegirine lujavrite I, the onset of macrorhythmic units and shifts in the chemistry of cumulus eudialyte are not integrated. Conceivably, a single pulse of crystallization could settle out as several macrorhyth­mic units reflecting the periodic arrival of a single crys­tal-charged current, but even here the step-like shift should occur at the end of the last deposited unit rat­her than in the middle of a unit as observed.

The origin of macrorhythmic units in the lower

layered kakortokites at Ilimaussaq was considered by Larsen and Sørensen (1987) and the reader is referred to their discussion. Largely by analogy with recent ideas on liquid layering in magma chambers, they proposed that one or several complete macrorhythmic units fractionated from an individual double-diffusive liquid layer until density equalization occurred with the overlying liquid layer. If similar processes occurred in aegirine lujavrite I, then density equalization may have been delayed as a result of the lower bulk density of these cumulates relative to the kakortokites. Further­more, because the bulk density of the aegirine lujav­rite I macro- and cryptorhythmic units in drill core 7 decreased with time, an increasingly delayed equaliza­tion of density would be expected and might explain the tendency for more macrorhythmic units to occur in the upper cryptorhythmic units of the drill core.

In the case of aegirine lujavrite I, the model of Lar­sen and Sørensen (1987) fails to explain why fractional crystallization within a macrorhythmic unit did not lead to shifts in the composition of cumulus eudialyte in the succeeding macrorhythmic unit. Contents of Y in cumulus eudialyte are apparently constant even in a sequence of four macrorhythmic units within cryptor­hythmic unit C4.

As an alternative, solidification of a liquid layer may have preceded not by fractional crystallization but by an extreme case of in situ crystallization during which residual magma escaped in such small amounts from the crystallizing cumulates that the composition of the liquid layer did not detectably evolve (Langmuir 1989).

In situ crystallization seems intuitively compatible with several features of the macrorhythmic units in aegirine lujavrite I: the subdued modal variations, the moderate to poor density sorting and the absence of a sharp onset to each macrorhythmic layer. It could proceed by a process of oscillatory nucleation (Maaløe 1978; Hort et al. 1993), though McBirney and Noyes (1979) argue that such nucleation occurs over scales of a few cm up to one m rather than over thicknesses of 6-17 m as found in macrorhythmic layers in aegirine lujavrite I. Furthermore, the crystallization is not truly oscillatory since, after crystallization of the eudialyte, microcline and nepheline layer, there is development of a cumulate layer relatively rich in intercumulus material rather than an immediate return to aegirine crystallization. An extreme, non-fractionating variety of in situ crystallization, of course, would fail to gene­rate a residual, less dense magma and another mecha­nism must be appealed to in order to generate the succession of apparently lighter macrorhythmic units in aegirine lujavrite I.

It is conceivable that in situ crystallization can be reconciled with the formation of macrorhythmic units by varying nucleation and growth rates. As Parsons and Becker (1987) emphasise for the Klokken intru­sion, physicochemical controls on nucleation and growth rates are poorly understood and no attempt is made here to extend their calculations to the more

12 Bulletin of the Geological Society of Denmark

ZdUU

1800

1400

Y ppm

10001

600

-i 1 r i 1 1 1 r

•18-5-87-5 m r e p e a t e d by f a u l t i n g „ 0-5-17-5 m

•89-5-121-5m

'122-5-151-5 m

•152-5-173-5 m

174-5-197-5 m

200

40 U

ppm

30

20

0-5-17-5 «18-5-105 m

10 2000 5000 10000 •15000 20000

Zr ppm

Fig. 6. Zr-Y and Zr-U compositional arrays for whole-rock samples in core 7, aegirine lujavrite I. Individual arrays are restricted to specific depth intervals and were defined by least-squares fitting; they represent cryptorhythmic layers which are considered to indicate the presence of liquid layers in the Ilimaussaq magma chamber.

esoteric magmas and minerals at Ilimaussaq. One should note, however, that this type of mechanism approaches an in situ character when (a) crystalliza­tion takes place in a magma chamber or layer of small vertical dimension thus minimising both collective and differential crystal settling and (b) the minerals involved are relatively small and have only minor den­sity differences from the magma. The presence of liquid layers in aegirine lujavrite I, the smaller size of the lujavrite cumulus crystals compared to both the underlying kakortokites and the Klokken syenites, and the relatively low density of eudialyte (2.88 g cm'3) all suggest that eudialyte crystallized more or less in situ. In agreement with this, eudialyte-rich layers do not underlie and cannot be separated from, as in the kakortokites, the layers rich in crystals of microcline

+ nepheline (s.g. 2.55 g.cm-3). Nevertheless, in situ crystallization does not explain the moderate density grading within macrorhythmic units and must be modi­fied to allow for some settling of crystals under gravity.

Development of cryptorhythmic layering The step-like geochemical shifts between composi-tionally distinct liquid layers may have been set up by sidewall crystallization leading to the rise of lighter, fractionated liquids towards the roof zone of the magma chamber (cf. McBirney 1980; Turner and Campbell 1986). This process would have been minimised at the kakortokitic stage of the Ilimaussaq intrusion when the intrusion had a large aspect ratio of 8-50 (Bailey et

Bailey: Cryptorhythmic and macrorhythmic layering 13

al. 1981a) but would have accelerated at the lujavritic stage when, after roof foundering, the Ilimaussaq chamber rapidly converted to two or more sub-cham­bers with lower aspect ratios. Chemically, these liquids should be enriched in Y and U relative to Zr since at this stage in the Ilimaussaq chamber Y and U have bulk partition coefficients <1 whereas Zr has a coeffi-cient>l (Andersenetal. 1981b).Furthermore,because fractionation takes place at an exceptionally evolved stage of the Ilimaussaq system (probably >95% crystallization) and in a sub-chamber of restricted volume, the rising sidewall liquids will show a large increase in their contents of incompatible elements (including Y and U) and these increases will be trans­ferred to the liquid layers and, on crystallization, to their cumulus eudialyte.

Most of the preceding discussion has assumed that the cryptorhythmic units in aegirine lujavrite I are best defined by shifts in the Y contents of eudialyte. Shifts in U contents only coincide with those for Y in one instance and seem to demand a different origin. Previ­ous workers (e.g. Steenfelt & Bohse 1975, Andersen et al. 1981b) have appealed to sudden losses of volatiles and complex ions such as UF6

2~ to explain unusual fea­tures in the distribution of U in Ilimaussaq materials. Similar processes, operating at the aegirine lujavrite I stage, may have influenced the transfer of U in sidewall liquids or perhaps from one liquid layer to the next.

Finally, one is left with a series of questions on why cryptorhythmic units always terminate and start in eudialyte-rich layers. Although eudialyte is a less dense mineral than aegirine, its chemical components are around 24% heavier than those of aegirine and it will thus be more effective in lowering the density of the host magma and triggering density equalization with the overlying liquid layer. This influence, however, is more than offset by the much greater abundance of cumulus aegirine. Is there a fixed delay between the period of aegirine-rich crystallization and the moment of density equalization in the succeeding eudialyte-rich layer? Does in situ oscillatory crystallization pro­ceed across cryptorhythmic boundaries regardless of the changes in major- and trace-element chemistry inferred between one liquid layer and the next? And why are there no obvious modal or petrographic changes when crystallization starts in a new liquid layer? Clearly there is considerable scope for further research on fundamental aspects of layering in the Ili­maussaq intrusion.

Tentative conclusions Cryptorhythmic layering has been identified by monito­ring the chemistry of cumulus eudialyte in a drill core through the earliest lujavrite (orthocumulate aegirine lujavrite I) of the Ilimaussaq intrusion and is taken as evidence for the presence of double-diffusive liquid layers. The appearance of cryptorhythmic layering is

linked to foundering of the intrusion's roof, the appea­rance of a sub-chamber with a sloping roof, a lower aspect ratio and a smaller vertical interval than the preceding kakortokitic stage when bottom cumulates probably formed over the full extent of the intrusion.

The double-diffusive liquid layers, with lower densi­ties but higher Y/Zr and U/Zr ratios at higher strati-graphic levels, may have been established by ascent of a low density fractionated magma along the sloping roof. The extreme fractionation of such agpaitic mag­mas promoted detectable cryptic changes in the liquid layers and their cumulus eudialyte.

In contrast, the internal crystallization of each crypto­rhythmic unit did not lead to internal cryptic changes in eudialyte chemistry suggesting that crystallization on or near the floor of the lujavrite sub-chamber occur­red by an extreme in situ, rather than a fractional, process.

Crystallization of each cryptorhythmic unit took place via the formation of two to four macrorhythmic units but the boundaries of the two types of layering do not coincide.

Ten macrorhythmic units have been recognised in the lujavrite drill core. Each cumulate unit consists of three sub-layers enriched in aegirine, then in eudialyte, microcline and nepheline, and finally in intercumulus phases (zeolites and arfvedsonite). They are charac­terized, relative to the macrorhythmic units of the earlier lower layered kakortokites, by subdued modal variations, moderate-to-poor density sorting and a diffusely marked start. It appears that these features are largely consistent with in situ crystallization though the moderate to poor density sorting requires some settling of crystals under gravity.

Acknowledgements I am grateful to the Geological Survey of Greenland and the Risø National Laboratory for access to drill core materials from the Ilimaussaq intrusion. Jens Henriksen kindly gave permission to reproduce Fig. 3. The support for the U analyses by the Risø National Laboratory and for XRFS analyses by the Danish Natural Science Research Council (SNF) is gratefully acknowledged. L.M. Larsen, H. Sørensen, J.R. Wilson and two anonymous referees made a critical and beneficial appraisal of the manuscript. Publication of the paper was authorized by the Geological Survey of Greenland.

Dansk sammendrag Kryptorytmisk lagdeling (Wilson 1992) er påvist i en 200 m lang borekerne gennem en lagdelt ægirin-eudialyt nefelin syenit, ægirin lujavrit I, fra den alkaline Ilimaussaq intrusion. Indholdet af Y i kumulus eudialyt

14 Bulletin of the Geological Society of Denmark

er konstant inden for hver kryptorytmisk enhed, men ændres brat til en højere værdi ved overgangen til en ny enhed. Hver kryptorytmisk enhed indeholder fra to til fire makrorytmiske enheder, der viser moderat til dårlig sortering efter vægtfylde,. og består af tre lag defineret ved successivt høje indhold af kumulus ægirin, fulgt af kumulus eudialyt, mikroklin og nefelin og endelig interkumulus zeoliter ;og arfvedsonit. Den "tørre" vægtfylde af både kumulus og interkumulus materialer aftager i de øvre kryptorhytmiske enheder, hvilket antages at afspejle aftagende vægtfylde "moder­væskens". Fremkomsten af en ny kryptorytmisk en­hed falder ikke sammen med afslutningen af en makro­rytmisk enhed og har tilsyneladende ingen relation til nukleation og bundfældning af kumulus faser. Kryst­allisation af kryptorytmiske enheder fra kemisk veldefi­nerede dobbelt diffusive væskelag i magmaet for ægirin lujavrit I og af makrorytmiske enheder ved gentagen in situ krystallisation på Ilimaussaq magmakammerets bund diskuteres.

References Andersen, S., Bohse, H. & Steenfelt, A. 1981a. A geo-

logicalsection through the southern part of the Ilimaussaq intrusion. Rapport Grønlands Geologiske Undersøgelse 103: 39-42.

Andersen, S., Bailey, J. C. & Bohse, H. 1981b. Zr-Y-U strati­graphy of the kakortokite-lujavrite sequence, southern Ilimaussaq intrusion. Rapport Grønlands Geologiske Un­dersøgelse 103: 69-76.

Bailey, J. C. & Gwozdz, R. 1994. Li distribution in aegirine lujavrite, Ilimaussaq alkaline intrusion, South Greenland: role of cumulus and postcumulus processes. Lithos 31: 207-225.

Bailey, J.C., Larsen, L.M. & Sørensen, H. 1981a. Introduc­tion to the Ilimaussaq intrusion with a summary of the reported investigations. Rapport Grønlands Geologiske Undersøgelse 103: 5-17.

Bailey, J. C , Bohse, H. & Demina, A. 1981b. Extension of Zr-REE-Nb resources at Kangerdluarssuk, Ilimaussaq in­trusion. Rapport Grønlands Geologiske Undersøgelse 103: 63-67.

Bailey, J. C , Rose-Hansen, J. & Sørensen, H. 1983. Leaching of uranium and thorium as a petrogenetic indicator in the agpaitic Ilimaussaq intrusion, South Greenland. In: Augu-stithis, S. S. (ed): The Significance of Trace Elements in Solving Petrogenetic Problems & Controversies, 861-885. Athens: Theophrastus Publications.

Bohse, H. & Andersen, S. 1981. Review of the stratigraphic divisions of the kakortokite and lujavrite in southern Ili­maussaq. Rapport Grønlands Geologiske Undersøgelse 103: 53-62.

Bohse, H., Brooks, C. K. & Kunzendorf, H. 1971. Field ob­servations on the kakortokites of the Ilimaussaq intrusion, South Greenland, including mapping and analyses by por­table X-ray fluorescence equipment for zirconium and nio­bium. Rapport Grønlands Geologiske Undersøgelse 38: 43 pp.

Bowen, N. L. 1915. Crystallization differentiation in silicate liquids. American Journal of Science 4th Series: 175-191.

Bowen, N. L. 1928. The Evolution of the Igneous Rocks. Princeton, N.J.: Princeton University Press.

Campbell, I. H. 1978. Some problems with the cumulus theory. Lithos 11: 311-323.

Ferguson, J. 1964. Geology of the Ilimaussaq alkaline intru­sion, South Greenland. Description of map and structure. Meddelelser om Grønland 172,4: 81 pp.

Ferguson, J. 1970. The significance of the kakortokite in the evolution of the Ilimaussaq intrusion, South Greenland. Meddelelser Grønland 190,1: 193 pp (Bulletin Grønlands Geologiske Undersøgelse 89).

Govindaraju, K. 1994. 1993 compilation of working values and sample descriptions for 383 geostandards. Geostan-dards Newsletter 18, Special Issue, July 1994, 1-158.

Helz, R. T. 1987. Differentiation behavior of Kilauea Iki lava lake, Kilauea Volcano, Hawaii: An overview of past and current work. In: Mysen, B. O. (ed): Magmatic Pro­cesses: Physicochemical Principles. The Geochemical Society, Special Publication No. 1: 241-258.

Henriksen, J. H. H. 1993. Geochemical trends through part of the upper transitional kakortokites and aegirine lujavrite I, Ilimaussaq intrusion, south Greenland. Unpublished M.Sc.thesis, Copenhagen University. 185 pp.

Hort, M., Marsh, B. D. & Spohn, T. 1993. Igneous layering through oscillatory nucleation and crystal settling in well-mixed magmas. Contributions to Mineralogy and Petro­logy 114: 425-440.

Irvine, T. N. 1987. Glossary of terms for layered intrusions. In: Parsons, I. (ed): Origins of Igneous Layering, 641-647. Dordrecht: D. Reidel.

Langmuir, C. H. 1989. Geochemical consequences of in situ crystallization. Nature 340: 199-205.

Larsen, L. M. & Sørensen, H. 1987. The Ilimaussaq intrusion - progressive crystallization and formation of layering in an agpaitic magma. In: Fitton, J. G. & Upton, B. G. J. (eds): Alkaline Igneous Rocks. Geological Society of Lon­don Special Publications No. 30: 473-488.

Maaløe, S. 1978. The origin of rhythmic layering. Minera-logical Magazine 42: 337-345.

McBirney, A. R. 1980. Mixing and unmixing of magmas. Journal of Volcanology and Geothermal Research 7: 357-371.

McBirney, A. R. & Noyes, R. M. 1979. Crystallization and layering of the Skaergaard intrusion. Journal of Petrology 20: 487-554.

Norrish, K. & Chappell, B.W. 1977. X-ray fluorescence spec­trometry. In: Zussman, J. (ed): Physical Methods in Deter­minative Mineralogy, 2nd Edition, 201-272. London: Academic Press.

Parsons, I. & Becker, S. M. 1987. Layering, compaction and postmagmatic processes in the Klokken intrusion. In: Parsons, I. (ed): Origins of Igneous Layering, 29-92. Dor­drecht: D. Reidel.

Sparks, R. S. J., Huppert, H. E. & Turner, J. S. 1984. The fluid dynamics of evolving magma chambers. Philoso­phical Transactions of the Royal Society of London, A 310: 511-534.

Steenfelt, A. & Bohse, H. 1975. Variations in the content of uranium in eudialyte from the differentiated alkaline Ili­maussaq intrusion, South Greenland. Lithos 8: 39-45.

Sørensen, H. 1969. Rhythmic igneous layering in peralkaline intrusions. Lithos 2: 261-283.

Bailey: Cryptorhythmic and macrorhythmic layering 15

Turner, J. S. & Campbell, I. H. 1986. Convection and mixing in magma chambers. Earth-Science Reviews 23: 255-352.

Turner, J. S. & Gustafson, L. B. 1981. Fluid motions and compositional gradients produced by crystallization or mel-ting at vertical boundaries. Journal of Volcanology and Geothermal Research 11: 93-125.

Ussing, N. V. 1912. Geology of the country around Juliane-haab, Greenland. Meddelelser om Grønland 38: 376 pp.

Vlasov, K. A. 1966. Mineralogy of Rare Elements, Vol. II: 355-364. Jerusalem: Israel Program for Scientific Trans­lations.

Wager, L. R. 1963. The mechanism of adcumulus growth in the layered series of the Skaergaard intrusion. Special Pa­per Mineralogicai Society of America 1: 1-9.

Wager, L. R. & Brown, G. M. 1968. Layered Igneous Rocks, 588 pp. Edinburgh: Oliver & Boyd.

Wilson, A. H. 1992. The geology of the Great Dyke, Zim­babwe: Crystallization, layering, and cumulate formation in the PI pyroxenite of cyclic unit 1 of the Darwendale Subchamber. Journal of Petrology 33: 611-663.

16 Bulletin of the Geological Society of Denmark